Weak deontic necessity in Japanese:
the case of beki
Lars Larm
（Senior Lecturer at the Department of Languages and Literatures, University of Gothenburg, Sweden）
松 山 大 学
言語文化研究 第 巻第 − 号（抜刷） 年 月
Matsuyama University Studies in Language and Literature
Weak deontic necessity in Japanese:
the case of beki
Lars Larm
＊．Introduction
）In Japanese, the markers of weak deontic necessity are the auxiliary beki ‘should’, as in ⑴, and the periphrastic hoo ga ii ‘had better’, as in ⑵, of which the former is stronger than the latter.） Strong deontic necessity is expressed by the
periphrastic nakereba naranai ‘must’, which is exemplified in ⑶.），）
＊ Senior Lecturer at the Department of Languages and Literatures, University of Gothenburg, Sweden
）An early version of this paper was presented at the International Modality Workshop held at Kansai Gaidai University in Osaka in . I wish to thank the participants for their support and encouragement. I am particularly grateful to Susumu Kubo, Harumi Sawada, Atsuko Wasa, Osamu Sawada, and Jun Sawada for useful suggestions and remarks. I also presented the topic at the seminar of the Lund Circle of East Asian Linguistics, in April , and would like to thank the audience for a stimulating discussion. I would also like to express my gratitude to Yukinori Takubo and Axel Svahn for valuable comments and corrections to the draft of this paper.
）The formal nouns mono ‘thing’ and koto ‘fact’ can also be used deontically（see Murata : − and − , and Narrog : ）. Makino and Tsutsui（ : ）compare beki with
mono, and explain that the two markers differ in that “mono da is used only in a generic
statement to express a social norm”. As their examples below show, it is not used to refer to a “specific situation” :
Gakusei wa benkyoo suru mono da.
students TOP study do. NPAST thing COP. NPAST ‘Students should study.’
Kimi wa benkyoo suru beki / *mono da.
you TOP study do. NPAST MOBL / thing COP. NPAST ‘You should study.’
⑴ Kimi wa sugu ik-u beki da.
you TOP immediately go-NPAST MOBL COP. NPAST ‘You should go immediately.’
⑵ Tabe-ta hoo ga i-i.
eat-PAST direction NOM good-NPAST ‘（You）had better eat.’
⑶ Tabe-na-kereba nara-na-i.
eat-NEG-COND become-NEG-NPAST ‘（You）must eat.’
‘（lit.）Not eating won’t do.’
The focus of this paper is beki ‘should’. In my discussion of the properties pertaining to this marker, I shall also, where appropriate, draw on insights from research on deontic modals in English.
Let me start with a general remark on the difficulty of describing the meaning and use of a grammaticalized modal marker. Leech（ : ）comments on this problem in his discussion of the modal auxiliaries in English :
Many pages, chapters, even books, have been written about the modal auxiliary verbs in English. What makes it so difficult to account for the use of these words（which may be called ‘modal auxiliaries’ or ‘modals’ for short）is that their meaning has both a logical and a ）The following abbreviations are used : ACC = accusative, ADN = adnominal, ASSUM = assumptive, COMP = complementiser, COND = conditional, CONJ = conjectural, COP = copula, DAT = dative, DED = deductive, EVID = evidential, FOC = focus particle, GEN = genitive, GER = gerund, HEAR = hearsay, INFIN = infinitive, LIT = literary, MOBL = moral obligative, NEG = negative, NML = nominaliser, NOM = nominative, NPAST = nonpast tense, PASS = passive, PAST = past tense, POL = polite, PRO = pronominal, QP = question particle, SPEC = speculative, TOP = topic.
） nakereba naranai ‘must’ has the variant form nakute wa ikenai, which in colloquial and informal speech can be contracted to nakucha.
practical（or pragmatic）element.
In the years that have passed since Leech wrote this passage, the literature on modality has mushroomed immensely with publications from various theoretical perspectives and with investigations presenting cross-linguistic data. The research literature on Japanese modality is also extensive, both within the indigenous Japanese grammatical tradition and within general theoretical frameworks （see
Sawada , Larm , , Narrog and the references therein）.
However, in spite of this amount of research, Leech’s remark on the ‘logical’ and ‘practical（pragmatic）’ aspects of modals is still relevant. He goes on to say
（Leech : − ）:
We can talk about them［the modal auxiliary verbs］in terms of such logical notions as ‘permission’ and ‘necessity’, but this done, we still have to consider ways in which these notions become remoulded by the psychological pressures which influence everyday communication between human beings: factors such as condescension, politeness, tact, and irony. Condescension, for example, intervenes to make the may of You may go（which in logical terms means no more than ‘permission’）into something approaching a command. As we will see during the course of this paper, beki, too, can be viewed from different angles. Its meaning has been described as having to do with moral obligation or “moral expectation”（Alfonso : ）, and it can be translated as ‘ought to’, ‘should’, or ‘supposed to’. As Johnson（ : ）succinctly states, “the feeling is that a judgment is being made based on social expectations”. Furthermore, some aspects of the meaning of beki can be couched in terms of the well-known dichotomies of root/non-root, weakness/strength, agentivity/non-agentivity, and objectivity/subjectivity. Although all four dichotomies will be given due attention in this paper, the issue of subjectivity is most prominent. Scholars
differ in their positions on this point. Kindaichi（ ）, Nomura（ ）and Larm （ ）characterize beki as an objective modal marker whereas Takanashi（ ）
takes the view that it is subjective when used sentence-finally. I this paper, I suggest both positions may be right depending on whether subjectivity is conceived of grammatically or pragmatically.
The structure of this paper is as follows. In the next section, I briefly touch upon the historical development of beki. In section , I discuss its form and distribution in present-day Japanese and also bring attention to two possible ongoing changes. Next, in section , I consider the semantic properties of beki both from the perspective of other scholars’ observations of weak necessity modals in English and from a language-specific point of view. I then discuss the issue of subjectivity, in section , which is followed by some concluding remarks.
．Historical development
Although the focus of this paper is beki in present-day Japanese, it is appropriate here to include a few remarks about its origin. In Old Japanese there was a modal, be-, which is one of the markers that Frellesvig（ : ）labels “verb extensions”. He explains that they “are inflecting clitics” and that “they follow a finite verb form to form an extended verb syntagm”（Frellesvig : ）. Be- was inflected as an adjective and had the conclusive form beshi, the attributive form beki, and the infinitive form beku.
The polysemous nature of be-, with both epistemic and deontic uses, and its semantic development have been of interest for scholars of historical linguistics and grammaticalization.） Narrog（ : ） explains that “be- gradually vanished
）Narrog（ ）provides a detailed investigation into the range of meanings expressed by the Old Japanese beshi, and Narrog（ , chapter ）discusses its historical development.
from the spoken language of western Japan, i. e. from those dialects of Japanese on which most of the written material in Middle Japanese is based”. It did, however, continued to be used in writing. Furthermore, be- has been preserved in the form of bee in many of the dialects of eastern Japan（see Narrog , chapter ）. For example, Fäldt（ ）, making reference to research by Gen Tamakake, presents examples from the Sendai City dialect illustrating hortative and conjectural uses of bee.
As for beki in present-day standard Japanese, Narrog（ : ）importantly points out that it “is a recent reintroduction of the marker from written language, not a historical continuation of the Old Japanese marker.” This happened at some stage during the Meiji Period（ − ）. The modern form beki is used both in adnominal and clause-final position, and it is specialized for expressing deontic meaning（although there may be an epistemic use in adnominal position, as we shall see in section ）.
．The form and distribution of beki in present-day Japanese
In clause-final position, beki is accompanied by the predicative form of the copula, whose nonpast form is da with the polite version desu. When beki modifies a noun the copula is covert in the nonpast tense, but in the past tense it takes the form datta. Before the pronominal no and the explanatory noda the copula has the adnominal form na in the nonpast tense and datta in the past tense. These patterns are illustrated below.⑷ Ik-u beki da/desu.
go. NPAST MOBL COP. NPAST/COP. NPAST. POL ‘（You）should go.’
⑸ kuru beki Ø/dat-ta hito come. NPAST MOBL COP/COP-PAST person ‘the person who is/was supposed to come’
⑹ Juushi sare-ru beki na/dat-ta
importance do. PASS-NPAST MOBL COP. ADN/COP-PAST no ga anzensa no takasa desu.
PRO NOM safety GEN highness COP. NPAST. POL
‘A high level of safety should be given/should have been given importance.’
（adapted and abbreviated from the internet）
⑺ Ik-u beki na/dat-ta no da.
go-NPAST MOBL COP. ADN/COP-PAST NML COP. NPAST ‘（You）should go/should have gone.’
What, then, are the forms that beki can follow ? Observe first that it comes after the nonpast form of verbs, as in the above examples. However, in the case of the light verb suru ‘do’, for historical reasons, there is also the alternative form su beki, as illustrated below :）
⑻ Gakusei wa benkyoo suru/su beki da.
students TOP study do/do MOBL COP. NPAST ‘Students should study. ’
Beki does not occur after the past tense form. Soga（ : − ）explains that
it “requires the verb preceding it to be in the ru form” and that “the event described by the verb preceding this modality marker is always regarded as
）The su of su beki ‘should do’, in example ⑻, is a remnant conclusive form of the light verb
something ‘uncompleted’ or rather, as yet ‘unrealised’”. Consider the following ungrammatical example taken from Masuoka and Takubo（ : ）:
⑼ *Kimi wa kanojo to wakare-ta beki da.
you TOP she with split up-PAST MOBL COP. NPAST （Intended to mean）‘You should have split up with her.’
This restriction also applies to constructions with nouns or nominal adjectives going before beki（see also Adachi et al. : ）. Note that before nouns, nominal adjectives take the adnominal form of the copula, na, or the literary form of the copula de aru, as in benri na doogu/benri de aru doogu ‘a convenient tool’. As for nouns in prenominal position, the copula has the allomorphs no and de aru, as in gakusei no Hanako/gakusei de aru Hanako ‘Hanako, who is a student’. However, before beki only de aru, which provides the nonpast form, is possible :
⑽ Kyooiku wa kyooseiteki *na/de ar-u beki da.
education TOP compulsory COP. ADN/COP. LIT-NPAST MOBL COP. NPAST ‘Education should be compulsory.’
⑾ Sekai ga heiwa *no/de ar-u beki da.
world NOM peace COP. ADN/COP. LIT-NPAST MOBL COP. NPAST ‘There should be peace in the world.’
It should be pointed out that the two sentences above have a rather formal ring to them.
The situation with adjectival（i-adjectives）constructions is unstable because there may be an ongoing change, which I will return to after having presented the textbook account. The standard description is that beki, which needs a preceding verb in the nonpast form as shown above, cannot follow the i-form of adjectives, as in :
⑿ *Jibun ni kibishi-i beki da.
self with strict-NPAST MOBL COP. NPAST ‘One should be strict with oneself.’
In this respect, beki differs from clause-final modal markers such as the epistemic hazu ‘should’, ka mo shirenai ‘may’ and ni chigai nai ‘must’, and from the evidentials yoo da ‘it appears’ and rashii ‘it seems’. However, as pointed out by Adachi et al.（ : ）, instead of a construction such as that in ⑿, an adjective in the infinite form followed by aru ‘be’ can be used（aru in this position is characterized as a ‘dummy copula’ by Nishiyama ）. This amendment rescues the sentence from ungrammaticality.
⒀ Jibun ni kibishi-ku ar-u beki da.
self with strict-INFIN be-NPAST MOBL COP. NPAST ‘One should be strict with oneself.’
Makino and Tsutsui（ : ） also mention that ku aru beki, as in ⒀, is possible but “not commonly used”.
Consider now negation. From what was said above about ⑿, it follows that beki cannot take syntactic scope over a negated verb, which morphologically is adjectival. Most descriptions of Japanese modality that I have come across（with a few recent exceptions, which I will return to below）state that it is not possible to say something like（example taken from Adachi et al. : ）:
⒁ *Uso o tsuka-na-i beki da.
lie ACC tell-NEG-NPAST MOBL COP. NPAST （Intended to mean）‘One should not tell lies.’
On the other hand, the negated form beki dewa nai is perfectly grammatical, as in the following example from Moriyama（ : ）（see also Sawada : ）: ⒂ Taroo wa ashita ik-u beki de. wa. na-i.
Taroo TOP tomorrow go-NPAST MOBL COP. NEG-NPAST ‘Taroo should not go tomorrow.’
Moriyama points out that the external negation in this example is superficial, and that the expressed meaning is rather［［ikanai］beki］where the modal takes semantic scope over negation. Thus, semantically beki itself cannot be negated（Moriyama
: − ）.），）
As I hinted at previously, there may be a change in progress, or perhaps two different but related changes, regarding adjectives and negated verbs followed by beki. Let us start with the latter, that is, sentences with the sequence nai beki, as in ⒁ above. This construction has been considered grammatically incorrect but may occur in actual usage. The scholar who first brought this matter to my attention was Takashi Masuoka, who in pointed out that the construction may
）See Narrog（ : ）for a possible counterexample.
）Consider also Sawada’s（ : ）examples below. The negative polarity item nani mo is semantically associated with the negative form of the verb, not with the negation of beki.
*Ashita made nani mo tabe-ru. tomorrow until what FOC eat-NPAST
（Intended to mean）‘（I）will not eat anything until tomorrow.’ Ashita made nani mo tabe-na-i.
tomorrow until what FOC eat-NEG-NPAST ‘（I）will not eat anything until tomorrow.’
Ashita made nani mo tabe-ru beki de. na-i.
tomorrow until what FOC eat-NEG-NPAST MOBL COP. NEG-NPAST ‘（You）should not eat anything until tomorrow.’
*Ashita made nani mo tabe-ru beki da.
tomorrow until what FOC eat-NEG-NPAST MOBL COP. NPAST ‘（You）should not eat anything until tomorrow.’
be used by young people. Furthermore, Kato（ ）presents some examples from The Balanced Corpus of Contemporary Written Japanese（BCCWJ）and notes that it is possible that the use will increase. Narrog（ : ）, too, points out that “examples can be found on the internet”. There are also comments on the internet on the correctness of such examples. Particularly worth mentioning is that the linguist Takumi Tagawa reports on his blog that the issue was discussed at a research seminar held in .） He writes that the teaching staff seemed surprised
to hear that young speakers accept nai beki while the students did not consider it problematic.
As for the other possible change, that is, the use of beki after an adjective, information is scarce. However, Tagawa comments that many of the speakers who accept nai beki also accept sentences such as :
⒃ Kyooshi wa kashikoi beki da.
teacher TOP clever MOBL COP. NPAST ‘A teacher should be clever.’
Thus, it seems that the combination of a negated verb followed by beki has entered the language of young speakers, and possibly this is also the case for adjectives.
．Meaning
Let us consider the meaning of beki from the perspective of observations on English modals. In the literature a distinction is often drawn between ‘root’ and ‘propositional’ modality. In sentences with root modality the modal is said to have a thematic relation with the subject while a propositional modal takes scope over a
proposition. For example, in ⒄ ‘must’ is deontic and marks an obligation pertaining to Mary while the epistemic ‘must’ in ⒅ expresses certainty about the content ‘Mary is at home’.
⒄ Mary must pay her taxes before next week. = Mary is obliged to pay her taxes before next week.
⒅ Mary must be at home（because the lights are on）.
= It is certain that Mary is at home（because the lights are on）.
Root modals have been described as having control structure and propositional modals raising structure. I shall refrain from discussing whether the root/ propositional dichotomy is the most appropriate characterization of modal markers in general. It is, in any case, useful when describing beki, since, as Takubo（ : ）explains, beki can be described as a root modal in the sense described above （although there are exceptions, which will be discussed below）. Unlike English modals, Japanese modal markers are not ambiguous between epistemic and deontic readings, but the difference can be shown by comparing beki with hazu（which can be glossed, at least roughly, as epistemic ‘should’）.） The following sentences,
adapted from Takubo（ : ）, show that beki has a thematic relation with the ）Beki can be viewed as a deontic expectational and hazu as an epistemic expectational marker. According to Narrog（ : ）:
Hazu and beki are two modal markers that have both been traditionally defined in terms of
‘matter of course’（cf. the comparison of both forms by Nakahata（ ））, and they can both frequently（but not always）be translated as ‘should’.［ . . . ］the ‘matter of course’ interpretation can be adduced from the existence of a strong presupposition, or expectation on the part of the speaker. This expectation may rest on moral values or personal convictions in the case of deontic beki, and knowledge in the case of hazu. Behind these expectations, then, the realization of some state-of-affairs is conceived as a ‘matter of course’.
subject, which must be animate, while no such restriction applies to hazu.
⒆ Tanaka ga kuru hazu/beki da.
Tanaka NOM come. NPAST ASSUM/MOBL COP. NPAST ‘Tanaka should come.’
⒇ Ame ga fur-u hazu/??beki da.
rain NOM fall-NPAST ASSUM/MOBL COP. NPAST ‘It should rain.’
This holds for prototypical uses of beki, which express that the subject is morally obliged to carry out an action, but it is also important to consider the possibility of non-agentive uses. It is here worth drawing attention to the distinction between ‘ought-to-do’ and ‘ought-to-be’ statements in English, which has long been a topic of discussion in philosophy（see Chrisman and Schroeder and the references therein）. Note that my concern is not the epistemic use where ‘ought’, of course, takes scope over the proposition. The point relevant here is that deontic ‘ought’ can be used not only to express that someone ‘ought’ to do something, as in ‘You ought to go home now’, but also to convey that a state of affairs ‘ought’ to be the case. Consider the following sentences from Schroeder（ : ）, which are examples of what he calls the “evaluative sense of ‘ought’” :
There ought to be world peace. World peace ought to obtain. The meeting ought to start at noon.
It ought to be that Jim jams.
Schroeder（ : ）states :
None of these sentences is plausibly understood as expressing a relation between an agent and an action, but each is plausibly interpreted as making a broadly normative claim, in contrast to a merely epistemic claim about what is likely to be the case.
Chrisman（ : ） too, although his perspective is different from that of Schroeder, discusses ‘agential and non-agential senses’ of the deontic ‘ought’ and presents the following examples which “cannot be assigned control syntax since their syntactic subjects do not conform to thematic-role restrictions”（Chrisman
: ）:
Invitations ought to go out by post.（Wedding Advice） Rump roast ought to cook slowly.（Cooking Advice）
Americanized spelling ought to be eliminated.（Editorial Advice）
I suggest that this line of thinking can be applied to the Japanese beki and propose a distinction between agentive type beki and non-agentive type beki. In ⑴, here repeated as , beki has an agentive sense. Note, however, that non-agentive uses are possible too, as in ⑾, here repeated as with a minor modification, , and
, which do not require an animate subject. Kimi wa sugu ik-u beki da.
you TOP immediately go-NPAST MOBL COP. NPAST ‘You should go immediately.’
Sekai ga heiwa de. ar-u beki da.
world NOM peace COP. LIT-NPAST MOBL COP. NPAST ‘There should be peace in the world.’
Kagaku wa shinpuru de. ar-u beki da.
science TOP simple COP. LIT-NPAST MOBL COP. NPAST ‘Science should be simple.’
Hoochoo wa tsune. ni surudo-ku ar-u beki da.
kitchen knife TOP always sharp-INFIN be-NPAST MOBL COP. NPAST ‘The kitchen knife should always be kept sharp.’
The last three sentences are formal, but as for , as we saw in the previous section, there may be an ongoing change in the direction towards acceptance of more colloquial forms.
Let me now move on to another well-known distinction, namely that between ‘strong’ and ‘weak’ deontic necessity markers. For instance, in English, ‘must’ and ‘have to’ are strong, and ‘ought to’ and ‘should’ are weak. ） Sloman（ :
− ）remarks :
It is worth comparing statements about what ought to happen or be done with statements about what is obligatory, essential , or what must happen or be done. Where the former kind of statement says（what is a necessary condition for）what is best, or better than all alternatives, the latter picks out the only candidate（or a necessary condition for it）. For instance ‘If you want to get to London by noon, then you ought to go by train’ picks out the best means without excluding the possibility of others, whereas ‘If you want to get to London by noon
）See Rubinstein（ ） and von Fintel and Iatridou（ ） for detailed discussions. Rubinstein（ ）provides an overview of theoretical approaches, but I shall not pursue their applicability to beki here.
then you have to（must, will be obliged to etc.）go by train’ implies that no other means exists.
The difference in degree of necessity between ‘ought to’ and ‘must’ is neatly captured in the following example from von Fintel and Iatridou（ : ）:
Everybody ought to wash their hands ; employees must.
They also point out that “some evidence for this relative weakness［of for example ‘ought’］comes from the fact that ⑶［ below］is not a contradiction while the examples in ⑷［ below］are”（von Fintel and Iatridou : ）:
You ought to do the dishes but you don’t have to.
a. #You have to do the dishes but you don’t have to. b. #You must do the dishes but you don’t have to.
Moriyama（ : ） notes that similar examples with beki are possible, particularly when it co-occurs with the expressions honrai wa ‘originally’ and tatemai to shite wa （from tatemae which means ‘official stance’）. I have not found actual examples equivalent to the above English sentences, but shows the weakness of beki. I owe this example to Yukinori Takubo（personal communication）:
Honrai nara chokusetsu at-te hanas-u beki da ga, origin COND directly meet-GER talk-NPAST MOBL COP. NPAST but jikan ga na-i node meeru suru.
time NOM not. exist-NPAST because email do. NPAST
Accordingly, strong modals, such as ‘must’ and ‘have to’ in English and the Japanese nakereba naranai ‘must’, are used in, for example, rule or law-abiding situations, such as the scenario of a passport control at the airport. In such contexts they are more felicitous than weaker modals.
You must/have to show your passport.
?You should/ought to show your passport.
Pasupooto o mise-na-kereba nara-na-i.
passport ACC show-NEG-COND become-NEG-NPAST ‘（You）must show show（your）passport.’
?Pasupooto o mise-ru beki da.
passport ACC show-NPAST MOBL COP. NPAST ‘（You）should show（your）passport.’
Referring to Adachi et al.（ : ）, Narrog（ : ）notes that, “beki can only be used when the modal target has a choice of action, while-（a）nakereba naranai can be used both in cases when there is and is not a choice”.
．Subjectivity
In this section, I turn my attention to the extensively discussed distinction between subjective and objective modality, which is one aspect of the near descriptive ineffability of modals. Recall Leech’s statements on the ‘logical’ and ‘practical（pragmatic）’ aspect of modals that I cited in the introductory section, which have bearing on this topic. That is, the deontic may in Leech’s example You may go is not subjective in logical terms, but from a practical or pragmatic
point of view we may say that it can be used to express the attitude of the speaker. Lyons（ ）is well known for having pointed out that even epistemic modals can, depending on the context, be taken objectively or subjectively. However, this does not mean that subjectivity is always pragmatic. It must be emphasized that there are modals with encoded subjectivity, where the ‘I−here−now’ perspective is part of the semantics of the form.
Let us see how this relates to beki. Narrog（ : ）states, about hazu ‘should’（epistemic）and beki ‘should’（deontic）, that “［ . . . ］ the association of these markers with speaker expectations, based on knowledge or moral values, make hazu and beki the most ‘subjective’ among the set of deontic and epistemic markers”. A similar position is taken by Takanashi（ ）who takes the view that beki is subjective when used sentence-finally. On the other hand, Kindaichi（ ）, Nomura（ ）, and Larm（ ）characterize beki as an objective modal marker. Although Kindaichi does not focus on beki in his paper, the following citation shows that he considers the imperative e（ro）as the subjective version of objective deontic modal expressions such as beki and nakereba naranai（Kindaichi :
, my translation）:
A command expressed by the imperative form is truly subjective. It is the command of the speaker, and moreover, it can only express a present command. If one wants to express a command-like meaning objectively, one would probably have to say shinakereba naranai ［must do］or su beki da［should do］.
What underlie the two perspectives mentioned above are different conceptions of subjectivity, which Kaufmann and Tamura（forthcoming : ）label the ‘pragmatic versus lexico-grammatical conception’ of subjectivity. They state （boldface in original）:
Authors disagree specifically on whether the distinction is to be drawn between linguistic expressions（strict lexico-grammatical conception, e. g. Kindaichi , Larm , Langacker , ）or between occurrences of linguistic expressions（pragmatic conception, Traugott and Dasher , Lyons ）, a crucial point for semantic theory building.
The issue is quite complicated and I shall not go into all details here. ） At the risk
of oversimplifying, my standpoint, which I touched upon at the beginning of this section, can be illustrated with the following examples of epistemic modality.
Hanako ga kuru kanoosei ga taka-i. Hanako NOM come. NPAST possibility NOM high-NPAST ‘There is a high possibility that Hanako will come.’
Hanako ga kuru daroo. Hanako NOM come. NPAST CONJ ‘Hanako will probably come.’
In the noun kanoosei ‘possibility’ can but does not have to be used subjectively to express the speaker’s attitude, which means that the subjectivity is not inherent in the form. If a listener says ‘No, that’s wrong’ she or he disagrees with the propositional content of the utterance, in which kanoosei ‘possibility’ is included, as in ‘I do not agree that there is a high possibility that Hanako will come’. On the other hand, daroo is inherently subjective, and performative, which means that saying ‘No, that’s wrong’ would be to disagree with the content ‘Hanako will come’, not with the modal attitude. This performative, non-truth-functional nature
）See Verstraete（ ）, De Smet and Verstraete（ ）, Kaufmann and Tamura（forthcoming）, and Narrog（ : chapter ）for detailed discussions, and, for my own view, see Larm（ and ）.
of daroo is reflected in that it resists embedding（see Larm for examples）. Embedding and other criteria used for testing whether a given marker is subjective or objective have been well known for a long time, at least since the publication of Kindaichi（ ）. Thus, although matters are more complex in actual usage, I think a distinction between pragmatically expressed and linguistically encoded subjectivity is warranted.
My point here is that regarding whether beki is subjective or objective, both positions may be right depending on whether subjectivity is conceived of pragmatically or grammatically. In actual usage, an utterance such as Iku beki da ‘You should go’ would most likely be used in situations where the speaker gives advice or perhaps, in some contexts, even issues an order. In this sense, it is correct to label it ‘subjective’. However, distributional facts show that beki can be propositionalized（although it is not entirely consistent throughout the tests I will use below）and in this sense it is objective.
Let us first consider two aspects of beki’s syntactic behaviour that do not support my claim that it is objective, not even from a grammatical point of view. These are scopal relations with negation and past tense. As we saw in ⒂ above, negation can follow beki, but as was noted, the embedding is syntactic rather than semantic. The same can be said about past tense, which only syntactically, but not semantically, can take scope over the modality. Adachi et al.（ : ）state that beki datta conveys the feeling of ‘regret’ or ‘dissatisfaction’. If my supposition is correct, ‘regret’ is expressed when the subject is first person and ‘dissatisfaction’ when it is non-first person, as exemplified below :
Boku wa ik-u beki dat-ta. I TOP go-NPAST MOBL COP-PAST ‘I should have gone.’
Kimi wa ik-u beki dat-ta. you TOP go-NPAST MOBL COP-PAST ‘You should have gone.’
Consider also the following example with the auxiliary verb oku :
Kippu o kat-te ok-u beki dat-ta. ticket ACC buy-GER do. in. advance-NPAST MOBL COP-PAST ‘I should have bought it（when I had the chance）.’
Narrog（ : ） remarks, about a similar example, that “［ . . . ］ no actual obligation in the past is reported. Instead, beki expresses the judgment of the speaker at time of speech that some event in the past［ . . . ］should have been realized.” Furthermore, in Narrog（ : ）, he explains that most examples with the past tense form beki datta carry an implication of “non-realization of the state-of-affairs”. He illustrates this with the following example（Narrog :
）:
??Kinoo wa, shigoto o suru beki dat-ta. yesterday TOP work ACC do. NPAST MOBL COP-PAST Dakara, shigotoba e it-ta.
therefore work-place to go-PAST
‘I should have worked yesterday. Therefore, I went to my workplace.’
Adachi et al.（ : ）also note the past negative form de wa nakatta and that the expressed meanings in such examples, too, are ‘regret’ and ‘dissatisfaction’. It seems that person is relevant as well, as we saw in examples and above. Observe the sentences below :
Boku wa ik-u beki de. wa. nakat-ta. I TOP go-NPAST MOBL COP. NEG-PAST ‘I shouldn’t have gone.’
Kimi wa ik-u beki de. wa. nakat-ta. you TOP go-NPAST MOBL COP. NEG-PAST ‘You shouldn’t have gone.’
However, other grammatical tests show that the modal source is not necessarily the speaker. First, there is the issue of interrogation, and, again, we may draw a parallel with the English may. In interrogative sentences, as Lehmann （forthcoming : ）puts it, “it is often systematically the hearer who is credited with the modal attitude.” He exemplifies by saying that “Linda may go is commonly interpreted as ‘I allow Linda to go’. However, the interrogative version May Linda go ? usually means ‘Do you allow Linda to go ?’ rather than ‘Do I allow Linda to go ?’”. This can be applied to beki, which can be questioned, as in the following example where the speaker, to borrow the wording of Lehmann（forthcoming : ）, “cedes the decision on the pragmatic focus to the hearer” :
Kare mo sanka suru beki desu ka. he too attendance do. NPAST MOBL COP. POL. NPAST QP ‘Is it his duty to attend as well ?’
Second, let us consider another subjectivity test, which is often used in the Japanese literature on modality, namely, embedding in relative clauses. A modal with encoded subjectivity, for instance the epistemic daroo, resists occurring in such embedded environments（see Larm for more details and examples）. But beki can be placed in adnominal position, as in and :
kuru beki Ø hito come. NPAST MOBL COP person ‘the person who is supposed to come’
Juushi sare-ru beki na importance do. PASS-NPAST MOBL COP. ADN no ga anzensa no takasa desu.
PRO NOM safety GEN highness COP. NPAST. POL ‘A high level of safety should be given importance.’
As noted previously, Takanashi（ ）takes the view that beki is subjective when used sentence-finally, but he also comments on and exemplifies adnominalisation. He makes the valuable observation that beki in this position, in addition to deontic readings, can also be used to express logical necessity as in the following example where the meaning is similar to that of epistemic hazu ‘should’ :
Moo tsui-te i-ru hazu no/ beki Ø already arrive-GER be-NPAST ASSUM COP. ADN MOBL COP kare ga mada araware-na-i.
he NOM still appear-NEG-NPAST
‘He, who is supposed to have arrived by now, is yet to turn up.’（Takanashi : ） Takanashi（ : − ） states that beki may have inherited this use from its predecessor be-, which, as I mentioned in section , was polysemous and had the adnominal form beki. This point is also noted by Moriya and Horie（ : ） who comment that beki “preserves its ambiguity in some contexts. For example, kuru beki hito either means ‘a person who（morally）should come’ or ‘a person who is（epistemically）supposed to come’.” However, it is nevertheless plausible that adnominalisation can be used as a subjectivity test, since also the present-day
deontic beki can be embedded.
The third point demonstrating the objective character of beki concerns combinations with other modal expressions. Nitta（ : ）reports that beki can be followed by the conjectural daroo and the speculative ka mo shirenai. This is confirmed by corpus data presented by Narrog（ ）（Narrog also adds hazu to the list of epistemic modals that may come after beki）. The data thus reflect the cross-linguistic pattern that epistemic modality belongs to a higher layer taking scope over deontic modality. Consider the following examples with the conjectural daroo and the speculative ka mo shirenai :
Ken ga ik-u beki Ø daroo.
Ken NOM go-NPAST MOBL COP. NPAST CONJ ‘I think that Ken should go.’
Ken ga ik-u beki Ø ka. mo. shirenai. Ken NOM go-NPAST MOBL COP. NPAST SPEC ‘Maybe Ken should go.’
The sense of these sentences is that Ken is seen as the most suitable or appropriate person to go. Note, however, that a different situation obtains with the epistemic ni chigai nai ‘must’. Nitta（ : ）states that he has not found any examples where this marker follows beki, and that these two probably cannot be combined. The same observation is made by Sugimura（ : ）. Thus, the following sentence would be regarded as unacceptable : ）
）However, Harumi Sawada（personal communication）points out that it might be possible to say, when driving on a highway :
Ichijikan. goto. ni kyuukei su beki ni. chigai. na-i. every. hour break do MOBL DED-NPAST ‘Certainly one should take a break every hour.’
*Ken ga ik-u beki Ø ni. chigai. na-i. Ken NOM go-NPAST MOBL COP. NPAST DED-NPAST （Intended to mean）‘Ken must be the one who should go.’
The data are less conclusive regarding evidential markers. Narrog（ : ） states that beki “resists embedding in evidentials”. However, in certain contexts it may be possible to place the hearsay soo da or evidential rashii after beki, as in :
Kimi ga ik-u beki da soo da.
you NOM go-NPAST MOBL COP. NPAST HEAR COP. NPAST ‘I hear that you should go.’
Kono kusuri o mainichi nom-u beki Ø rashii. ）
this medicine ACC every day drink-NPAST MOBL COP. NPAST EVID ‘It seems（you）should take this medicine every day.’
Observe also the following sentences from the internet, in which the evidential yoo da follows beki.
Kaminari ni chuui su beki Ø yoo desu.
thunder DAT caution do MOBL COP. NPAST EVID COP. NPAST. POL ‘It seems that（we）should be careful of the thunder.’
Moo sukoshi hayaku tanemaki o suru more a little early sowing ACC do. NPAST beki dat-ta yoo desu.
MOBL COP-PAST EVID COP. NPAST. POL
‘It seems that（I）should have planted the seeds a little earlier.’
）I am grateful to the participants of the International Modality Workshop, Kansai Gaidai University, , for the dialogue that led to this example.
Fourth, I shall briefly remark upon the propositional attitude verb omou ‘think’. Narrog（ : ）, making reference to Natsu Ozaki, explains that beki followed by to omou ‘I think’ is a harmonic construction. That is, the propositional attitude verb and beki co-express subjectivity. However, Susumu Kubo（personal communication）points out that beki can also be followed by the stative version omotte iru ‘think’, which is an objective marker（see Nakau ）, and, in Susumu Kubo’s wording, expresses ‘an intentional state’. Consider the difference between the sentences below. In the modal attitude holder is Taroo, not the speaker.
Ken ga ik-u beki da to omo-u.
Ken NOM go-NPAST MOBL COP. NPAST COMP think-NPAST ‘I think that Ken should go.’
Taroo wa Ken ga ik-u beki da to omot-te i-ru. Taroo TOP Ken NOM go-NPAST MOBL COP. NPAST COMP think-GER be-NPAST ‘Taroo thinks that Ken should go.’
Thus, the data indicate that beki can become part of the propositional content, and it is, from a strictly grammatical perspective, objective.
．Concluding remarks
I shall conclude this paper by suggesting avenues of further investigation. In section , on the morphological and distributional behaviour of beki, I drew attention to two changes that may be occurring. It would be valuable to systematically test these constructions against native speakers of different age groups. Further, when discussing the semantic properties of beki, in section , I restricted my focus to aspects that are amenable to investigation from the perspective of
the well-known dichotomies of root/non-root, weakness/strength, and agentivity/ non-agentivity. And, my treatment of subjectivity, in section , was limited to distributional tests aimed at determining whether or not beki, from a grammatical point of view, is subjective. However, beki also displays interesting language-specific characteristics in terms of its pragmatic and sociolinguistic properties. The functional and stylistic aspects of beki, and its use in different registers, merit further attention.
References
Adachi, T. et al.（Nihongo Kijutsu Bunpoo Kenkyuukai［The Society of Descriptive Japanese Grammar］）（eds）.（ ）. Gendai Nihongo Bunpoo : Dai Bu Modariti［Modern Japanese
Grammar : Part Modality］. Tokyo : Kuroshio.
Alfonso, A.（ ）. Japanese language patterns : a structural approach（vol. and ）. Tokyo : Sophia University L. L. Center of Applied Linguistics.
Chrisman, M.（ ）. ‘Ought’ and control. Australasian Journal of Philosophy. （ ）: − .
De Smet, H. and Verstraete, J-C.（ ）. Coming to terms with subjectivity. Cognitive Linguistics （ ）: − .
von Fintel K. and Iatridou S.（ ）. How to say ought in foreign : The composition of weak necessity modals. In : Guéron J. and Lecarme J.（eds）Time and modality. Studies in natural
language and linguistic theory, vol . Dordrecht : Springer. − .
Frellesvig, B.（ ）. A history of the Japanese language. Cambridge University Press. Fäldt, K.（ ）. The conjectural and hortative bee : From dialectal particle to stereotypic
marker. Bachelor’s thesis. Lund University.
Johnson, Y.（ ）. Modality and the Japanese language（Michigan Monograph Series in Japanese Studies ）. Ann Arbor : Center for Japanese Studies, The University of Michigan. Kato, E.（ ）. Koopasu ni motozuku ‘beki da’ no bunseki［A corpus-based analysis of beki
da］. The Bulletin of the Center for International Students at Asahi University. （ ）: − . Kaufmann, M. and Tamura, S.（Forthcoming）. Japanese modality−Possibility and necessity :
Prioritizing, epistemic, and dynamic. Preliminary draft of chapter to appear in Jacobsen, W. and Takubo, Y.（eds）. The Handbook of Japanese Semantics and Pragmatics. Moulton de Gruyter. Version : January , .
Kindaichi, H.（ ）. Fuhenka jodooshi no honshitsu−shukanteki hyoogen to kyakkanteki hyoogen no betsu nitsuite［The nature of non-inflectional auxiliaries : on the distinction between subjective and objective expressions］. Kokugo Kokubun［National Language and Linguistics］ （ − ）. Reprinted in : Hattori, S. et al.（eds）, Nihon no Gengogaku［Linguistics in Japan］, （vol. ）. Tokyo : Taishuukan. − .
Larm, L.（ ）. Modality in Japanese. DPhil thesis. University of Oxford.
Larm, L.（ ）. West meets East : A Kindaichian approach to subjective modality. In Pizziconi, B. and Kizu, M.（eds）, Japanese modality : exploring its scope and interpretation. Palgrave Macmillan. − .
Larm, L.（ ）. Modality packaging in Japanese : The encoding of modal meanings and subjectivity. Studies in Pragmatics : − .
Leech, G. N.（ ）. Meaning and the English verb. Burnt Mill, Harlow, Essex : Longman Group Limited.
Lehmann, C.（Forthcoming）. Speech-act participants in modality. Downloadable at https://www.christianlehmann.eu/publ/lehmann_modality.pdf
Accessed on August , .
Lyons, J.（ ）. Semantics（vol. ）. Cambridge : Cambridge University Press.
Makino, S. and Tsutsui, M.（ ）. A dictionary of intermediate Japanese grammar. Tokyo : The Japan Times.
Martin, S. E.（ ）. A reference grammar of Japanese. Rutland : Charles E. Tuttle Company. Masuoka, T. and Takubo, Y.（ ）. Kiso Nihongo Bunpoo ［Basic Japanese Grammar］（Rev.
ed.）. Tokyo : Kuroshio.
Moriya, T. & Horie, K.（ ）. What is and is not language-specific about the Japanese modal system ? A comparative and historical perspective. In Pizziconi, B. and Kizu, M.（eds）,
Japanese modality : exploring its scope and interpretation. Palgrave Macmillan. − . Moriyama, T.（ ）. Kihon johoo to sentaku kankei to shite no modariti［Modality as basic
moods and selectional relations］. In Moriyama, T., Nitta, Y., and Kudo, H. Modariti（Nihongo no bunpoo ）［Modality（Japanese Grammar ）］. Tokyo : Iwanami Shoten. − .
Murata, M.（ ）. Syntax and semantics of the nominals ‘mono’ and ‘koto’ in Japanese. Master’s thesis. Massey University.
Nakau, M.（ ）. Modariti to meidai［Modality and proposition］. In Hayashi Eiichi Kyooju Kanreki Kinen Ronbunshuu Kankoo Iinkai（eds）. Eigo to Nihongo to : Hayashi Eiichi Kyooju
Kanreki Kinen Ronbunshuu Kankoo［English and Japanese : A Publication of Collected Papers
for the th Birthday of Professor Eiichi Hayashi］. Tokyo : Kuroshio. − .
Narrog, H.（ ）. Polysemy and indeterminacy in modal markers−the case of Japanese beshi.
Narrog, H.（ ）. Modality in Japanese : The layered structure of the clause and hierarchies of
functional categories. John Benjamins Publishing Company.
Narrog, H.（ ）. The order of meaningful elements in the Japanese verbal complex.
Morphology : − .
Narrog, H.（ ）. Modality, subjectivity, and semantic change : A cross-linguistic perspective. Oxford : Oxford University Press.
Nishiyama, K.（ ）. The morphosyntax and morphophonology of Japanese predicates. Doctoral dissertation. Cornell University.
Nitta, Y.（ ）. Ninshiki no modariti to sono shuuhen［Epistemic modality and its surroundings］. In Moriyama, T., Nitta, Y., and Kudo, H. Modariti（Nihongo no bunpoo ）［Modality（Japanese Grammar ）］. Tokyo : Iwanami Shoten. − .
Nomura, T.（ ）. Modariti keishiki no bunrui［On the classification of Japanese modal forms］.
Kokugogaku［National Language Studies］ （ ）: − .
Rubinstein, A.（ ）. Roots of modality. Doctoral dissertation. University of Massachusetts Amherst.
Sawada, H.（ ）. Modariti［Modality］. Tokyo : Kaitakusha.
Sawada, H.（ ）. Ninshikiteki modariti to shookoteki modariti no aida［Between epistemic and evidential modality］. Proceedings of the International Modality Workshop［funded］via Grant-in-Aid for Scientific Research : − .
Schroeder, M.（ ）. Ought, agents, and actions. Philosophical Review. Vol. （ ）: − . Sloman, A.（ ）. ‘Ought’ and ‘better’. Mind . （ ）: − .
Soga, M.（ ）. Tense and aspect in modern colloquial Japanese. Vancouver : University of British Colombia Press.
Sugimura, Y.（ ）. Tsuzuki : kamoshirenai to nichigainai no ishitsusei : koopasu no kekka kara ［Sequel : the heterogeneous character of ni chigainai and kamoshirenai : from the results of the
corpus］. Kotoba to Bunka［Language and Culture］ : − .
Takanashi, S.（ ）. Hyooka no modariti o arawasu jodooshi beki da［Beki da : a modal expressing evaluation］. The Bulletin of the International Student Center, Kobe University. Vol. : − .
Takubo, Y.（ ）. Conditional modality : Two types of modal auxiliaries in Japanese. In Pizziconi, B. and Kizu, M.（eds）, Japanese modality : exploring its scope and interpretation. Palgrave Macmillan. − .
Verstraete, J-C.（ ）. Subjective and objective modality : Interpersonal and ideational functions in the English modal auxiliary system. Journal of Pragmatics : − .
